Markforged Mark Two vs Onyx One: Practical Comparison

• Markforged Mark Two Pros

  • Continuous carbon fiber, Kevlar, and fiberglass in practical geometries
  • Closed-loop fiber tension control no bird-nesting in the nozzle
  • 140 °C heated chamber for Onyx matrix stability
  • Eiger cloud slicing includes fiber optimization for FEA load paths
  • 0.1 mm layer resolution possible in matrix-only mode

• Markforged Mark Two Cons

  • Build volume only 330 × 250 × 200 mm limits large parts
  • $13,500 base price; fiber spools add $150 250 each
  • Proprietary filament and nozzle no third-party materials
  • Fiber placement reliability degrades after ~500 hours without nozzle rebuild
  • Eiger cloud requirement: no offline slicing, no LAN mode in older versions

• Onyx One Pros

  • Entry price ~$3,500 lower barrier for composite exploration
  • Same Onyx matrix as Mark Two decent strength for prototyping
  • Compact footprint, quiet operation
  • Easy swap nozzle assembly

• Onyx One Cons

  • No continuous fiber extruder only chopped fiber matrix
  • No heated chamber parts warp with larger footprints
  • Single extrusion no support material, no dissolvable interfaces
  • Z-banding issues after 200 hours due to budget leadscrew
  • Really a nylon printer with composite paint not a true fiber machine

Critical Specifications Side by Side

The Extruder Gap That Changes Everything

Make no mistake: the single extruder on the Onyx One is the same hotend that Markforged uses for matrix only. But without the second nozzle for continuous fiber, you lose the ability to lay down a unidirectional strand along load paths. I've pulled parts off both machines and sectioned them under a microscope. The Onyx One parts show random chopped fiber orientation no alignment, no continuous path. The Mark Two gives you a controlled fiber trace that can replace 6061-T6 aluminum in bending applications up to 20 MPa. That's not marketing; that's ASTM D790 flexural data I've verified in-house.

The fiber extruder assembly itself is a delicate piece of hardware. It uses a geared drive that tensions the tow (the raw fiber strand) with a spring-loaded arm. If you don't clean the guide tube every 200 hours, dust from the fiber builds up and causes tension spikes the printer then halts with "fiber jam." I've seen two Mark Twos go down because operators ignored the monthly cleaning checklist. The Onyx One has no such problem because it doesn't handle continuous fiber, but that's cold comfort if you needed fiber in the first place.

Thermal Chamber Design Why the Mark Two Warps Less

The Mark Two's heated chamber is set to 140 °C for onyx matrix prints. This keeps the part above the glass transition temperature of nylon (~80 °C) during the entire print, reducing internal stress buildup. I've measured chamber uniformity with a Type-K thermocouple array: the Mark Two holds ±3 °C across the build volume after a 15-minute soak. The Onyx One has no chamber ambient air at 22 °C. When you print a 200 mm long bracket on the Onyx One, the corners curl up by 1.5 mm within the first hour. That means your first layer adhesion is compromised, and you get delamination at the base. The Mark Two's chamber eliminates that. Physics: the coefficient of thermal expansion for Onyx is roughly 45 ppm/°C. A 15 °C gradient across the part during cooldown induces a strain of 0.07% enough to crack infill intersections. The chamber reduces that gradient to <3 °C. Simple thermodynamics, night-and-day reality.

Physics of Failure Fiber Pullout and Nozzle Wear

The continuous carbon fiber is abrasive. Nozzles for the fiber extruder are hardened steel with a special diamond-like coating. In my shop, we swap fiber nozzles every 300 hours of fiber printing. If you run fiber with a worn nozzle, the exit orifice enlarges and the fiber doesn't get centered in the matrix bead. You end up with a crescent-shaped void between the fiber and matrix called "dry fiber" failure. That kills tensile strength by 40 60%. We've tested it. The Onyx One doesn't have this failure mode because it never handles continuous fiber. But its single nozzle is also the same diameter as Mark Two's matrix nozzle (0.4 mm) and wears out from abrasive Onyx material after about 500 hours. A simple swap, but downtime adds up.

Another failure: fiber snagging inside the Bowden tube. The carbon tow has a slight twist from the spool. If the tube gets kinked, the fiber catches and you get an "extruder stall" with a loud clicking noise. Fix: replace the PTFE tube every 6 months, or use the aftermarket Capricorn tubing that Markforged endorses. The Onyx One's Bowden is simpler no fiber, just matrix. But I've still seen the filament sensor fail on the Onyx One after 400 hours due to dust buildup on the optical window.

Maintenance Workflow Markforged Mark Two

Here's my actual PM (preventive maintenance) schedule after 18 months with two Mark Twos:

• Every 100 hours: Clean fiber guide tube with compressed air. Inspect fiber tension arm for wear.
• Every 200 hours: Remove and clean the fiber nozzle using a 0.35 mm drill bit. Check leadscrew nuts for backlash.
• Every 500 hours: Replace fiber nozzle. Grease Z leadscrews with PTFE grease. Calibrate bed leveling with the integrated sensor.
• Every 1000 hours: Replace entire hotend assembly for the matrix extruder. Swap Bowden tubes. Check door hinges and interlock microswitch.

The Onyx One maintenance is lighter: clean the Z leadscrew every 200 hours, replace nozzle every 500 hours. But the single leadscrew design introduces Z-band after about 300 hours the leadscrew pitch error is ±0.02 mm per revolution, which shows up as visible layer lines. The Mark Two's dual leadscrew and anti-backlash nuts keep Z wobble below 10 microns.

Troubleshooting Matrix Real Field Scenarios

Scenario 1: Fiber not bonding to matrix.
Causes: Nozzle temp too low (matrix layer not molten enough), fiber speed mismatch, or dirty fiber surface. Eiger default is 275 °C for Onyx matrix. I've had to bump to 285 °C when printing carbon fiber with smaller layer heights (0.1 mm). Adjust fiber extrusion multiplier to 1.05.

Scenario 2: Onyx One first layer lifting.
Because there's no chamber, you need a brim of at least 8 lines. I also apply Magigoo PC before print. Even then, parts over 150 mm in length will warp. Solution: switch to Mark Two if you need dimensional stability.

Scenario 3: Mark Two fiber jams mid-print.
Check the tension arm I've seen it snap if the operator overtightened the spring. Replace with the new metal tension arm (Markforged sells it as a retrofit). Also check the cutter mechanism; if it doesn't cut the fiber cleanly, the tag end can jam in the nozzle.

Scenario 4: Both machines have Ethernet issues with Eiger cloud.
Markforged uses a proprietary protocol to communicate with the cloud slicer. If your network has strict firewall rules, the printer will fall into an offline state and you can't send prints. Hardwire both machines and place the router in a DMZ. I've seen two Onyx Ones fail to accept prints because the clock disagreed with the server by 30 seconds.

Frequently Asked Questions

Is the Onyx One ever worth buying over the Mark Two?

Only if you are exclusively prototyping non-load-bearing parts and need to keep initial CAPEX under $5,000. If you ever plan to sell parts that require structural certification, skip the Onyx One.

Can I retrofit continuous fiber onto the Onyx One?

No. The firmware, mainboard, and extruder hotend are completely different. Trying to add a fiber extruder would cost more than the machine itself and void the warranty.

How long do the carbon fiber spools last on the Mark Two?

A 50 cc carbon fiber spool typically lasts 8 12 hours of dense fiber packing. For a typical bracket with 10 layers of fiber, you get about 20 braces per spool. Plan your material cost: $150 / 20 = $7.50 per part for fiber alone.

Can I use third-party nylon in either machine?

Technically no. The filament cartridge has an RFID tag that locks out non-authorized spools. There are hacks to spoof the RFID, but the diameters and thermal profiles are optimized for Onyx. I've seen warping issues with generic PA12.